Contract No.: M-7042-8-00-97-30 MPR Reference No.: 8573 Left Out of the Boom Economy: UI Recipients in the Late 1990s Executive Summary October 2001 Karen Needels Walter Corson Walter Nicholson Submitted to: U.S. Department of Labor Employment and Training Administration Office of Workforce Security 200 Constitution Ave, N.W., Room S-4231 Washington, DC 20210 Project Officer: Anissa Holm Submitted by: Mathematica Policy Research, Inc. P.O. Box 2393 Princeton, NJ 08543-2393 (609) 799-3535 Project Director: Walter Corson
EXECUTIVE SUMMARY Since 1935, the unemployment insurance (UI) system has provided limited income support for workers who lose their jobs through no fault of their own. This support, which often replaces 40 to 50 percent of lost weekly earnings, continues until the unemployed worker either becomes reemployed or reaches his or her limit and exhausts benefits. In 1998, the year of this study, two states provided benefits for a maximum of 30 weeks, and the rest did so for 26 weeks. Nine states provided 26 weeks to all recipients, while the potential durations of benefits in the remaining states varied depending on the pre-ui earnings of the recipients. In 1998, 7.3 million people began receiving UI, and approximately 2.3 million, or 32 percent, exhausted their benefits. This study examines the characteristics, labor market experiences, UI program experiences, and reemployment service receipt of UI recipients who began collecting UI benefits in 1998. One objective is to gauge the extent to which recent changes in the U.S. labor market may have affected the composition of UI recipients who exhaust benefits and to examine their postexhaustion labor market behavior. Another objective is to examine recipients experiences with the delivery of reemployment services and determine whether changes in the workforce development system have affected these experiences. The study uses individual-level survey data on a nationally representative sample of UI recipients to examine the characteristics of recipients and their labor market and program experiences and aggregate state-level data to examine trends in UI exhaustion rates over time. The findings from this 1998 survey sample are compared to those from an earlier survey of UI recipients who began collecting benefits in 1988. The economy was strong in both years, with low unemployment rates (the unemployment rate in 1998 was 4.5 percent, as compared to 5.5 percent in 1988). STUDY DESIGN AND IMPLEMENTATION The UI exhaustee study design for the survey called for the selection of nationally representative samples of UI exhaustees and nonexhaustees and the collection of UI program data and telephone survey data from these samples. Sample selection was a two-step process: 25 states were selected in the first step, and exhaustees and nonexhaustees were selected in the second step. The exhaustees and nonexhaustees were people who established a benefit year in 1998 and received at least one payment. Interviews were attempted with subsamples of the exhaustees and nonexhaustees. The interviews were conducted in English and Spanish during an approximately seven-month period from mid-july 2000 to mid-february 2001. Interviews were completed with 3,907 UI recipients: 1,864 exhaustees and 2,043 nonexhaustees. Interviewing occurred in two stages. In the initial 16-week fielding period, we used mail, telephone, and database locating methods to attempt to find and complete telephone interviews with members of this sample. People interviewed in the initial stage are nationally representative of UI exhaustees and nonexhaustees who can be contacted and interviewed by telephone within 16 weeks. In a second, more intensive stage in a random subset of 10 states, we continued our iii
attempts to interview sample members. We continued our mail, telephone, and database locating activities and added field staff to find sample members who had not responded to our initial interview attempts. We asked people we located to call our telephone center to complete interviews. Those interviewed through this extended fielding period are nationally representative of UI exhaustees and nonexhaustees who require intensive locating efforts. Hence, the final sample of completed interviews has two components: (1) an initial fielding component obtained from 25 states, and (2) an intensive fielding component obtained from 10 states. The overall survey completion rate was 63 percent (65 percent for nonexhaustees and 61 percent for exhaustees), which is similar to the rate for the earlier study of recipients in 1988. POTENTIAL CAUSES OF THE HIGH 1990s UI EXHAUSTION RATE This study examines the characteristics of UI recipients and the relationship between these characteristics and UI exhaustion rates in the 1990s. Prior research has found that, during the booming economy of the late 1990s, the unemployed exhibited longer unemployment spells and worse reemployment outcomes than has historically been the case during strong economic times. UI recipients, about one-third of the unemployed, also experienced longer UI spells and higher benefit exhaustion rates than historical experience would predict. For example, a simple model of the exhaustion rate that statistically controls for the unemployment rate accurately predicts the 1988 exhaustion rate to be 28 percent. However, it predicts a 1998 exhaustion rate of 25 percent, approximately 7 percentage points lower than what was actually observed. We focused on changes over time, using both recipient-level and aggregate data, to explore potential reasons for the recent high exhaustion rates. We find that: Changes between 1988 and 1998 in recipient characteristics are a strong factor in the increase in exhaustion rates. Key changes leading to an increase in the aggregate exhaustion rate were the aging of the recipient population, the increase in the percentage who are Hispanic, and the decreases in the proportions who were from manufacturing and who had definite recall dates. If the recipient population, labor market backgrounds, and UI program parameters were unchanged between 1988 and 1998, we estimate that exhaustion rates would be about four to five percentage points lower in 1998 than was the case. These changes in the UI recipient population mirror broader changes in the labor market. Well-documented trends, such as the aging of the baby boom generation, the decline in manufacturing employment, and the increase in service sector industries are influencing the composition of the labor force, resulting naturally in changes in who becomes a UI recipient. Although most changes in UI system parameters such as wage-replacement rates or average potential durations were modest, a decline in the average UI potential duration explains part of the high exhaustion rates. Although the effects of several UI system parameters were usually statistically insignificant in the aggregate data analysis, analysis of recipient-level data suggests that a decline in potential duration explains some of the higher exhaustion rates between 1988 and 1998. iv
The unemployment rate and the duration of unemployment significantly affect exhaustion rates. Each percentage point increase in the unemployment rate was estimated to add two to three percentage points to the exhaustion rate if other factors are held constant. Increasing unemployment durations during the 1990s, which explain about half of the atypically high exhaustion rates during the period, are likely to be symptomatic of other underlying changes affecting UI and labor market outcomes. UI EXHAUSTEES CHARACTERISTICS UI exhaustees represent a particularly important group of workers for policymakers. As a group, they have strong work histories and have demonstrated attachment to the labor market in the past. However, their long UI spells suggest that they face particular difficulty finding new jobs. Documenting who they are can help policymakers who administer the UI program and reemployment services serve them better. Despite changes over time in recipient characteristics, relationships between these characteristics and whether a UI recipient exhausts his or her benefits have not changed much. In examining who UI exhaustees are, we find: Exhaustees are more likely than nonexhaustees to belong to some demographic groups. Females and racial/ethnic minorities are disproportionately likely to exhaust benefits compared to other groups. Prior to their UI claims, exhaustees are slightly more likely than nonexhaustees to have had lower-paying, part-time jobs that did not provide fringe benefits. They are also more likely to have been in nonstandard work arrangements (such as temporary work or self-employment) and less likely to have been in a union or in the manufacturing sector. Compared to nonexhaustees, exhaustees were less likely to have been laid off and were less likely to have expected to return to their former employers. Exhaustees were more likely to have quit, been discharged, or lost their job for some other reason. Predictably, therefore, they were less likely to have expected recall and less likely to have been recalled. Because of exhaustees lower pre-ui earnings, they typically had lower weekly benefit amounts and potential durations than nonexhaustees. Not surprisingly, having a shorter potential duration increases the likelihood that a recipient will exhaust his or her benefits. LABOR MARKET EXPERIENCES The economic consequences of job loss for UI recipients depend both on the length of time they are unemployed and the quality of the jobs they eventually obtain. Although unemployed workers in recent years are more likely to be long-term unemployed, compared to what has been the case historically, some recipients become reemployed very quickly because they are recalled to their pre-ui jobs at the same earnings rate as before. Understanding the labor market v
experiences of UI recipients, particularly of exhaustees, can therefore help to assess the longterm consequences resulting from job loss. We find, generally, that labor market outcomes for 1998 recipients were worse than those for 1988 recipients. Compared to 1988 recipients, 1998 recipients took longer to find a job and were less likely to become reemployed. For example, a little over one-quarter (28 percent) of UI recipients in 1998 were reemployed in 10 or fewer weeks, compared to 40 percent in 1988. At one year after job loss, reemployment rates were 65 percent for 1998 recipients, compared to 81 percent in 1988. Ultimately, twice as many 1998 recipients as 1988 recipients remained without any post-ui job during the entire follow-up period (21 percent versus 10 percent, respectively). Overall, work search rates were slightly lower in the late 1990s than in the late 1980s, although recipients who searched reported doing so intensively. In both 1988 and 1998, exhaustees were more likely than nonexhaustees to have looked for work, and they were more likely to look for work when they were collecting UI benefits than after they exhausted them. However, about 7 percent of recipients in 1998 appeared not to have searched for work shortly after they started collecting UI benefits because of reasons that imply they were out of the labor force. This contrasts with about four percent for 1988 recipients. In both 1998 and 1988 many recipients post-ui jobs paid less than their pre-ui jobs. When 1998 exhaustees became reemployed, they averaged a 16 percent reduction in their earnings at their first post-ui job, which was comparable to the earnings losses for exhaustees 10 years earlier. Nonexhaustees in 1998 experienced an average seven percent reduction in earnings, compared to a one percent reduction in 1988. However, this difference for nonexhaustees was attributable to recipients at the tails of the distribution. Overall the distribution of earnings changes was similar in the two years. In addition, 1998 recipients were more likely to report having a job with an alternative employment arrangement (such as being a temporary worker, independent contractor, or leased employee) after their UI spells than was the case prior to the UI claim. This is especially true for exhaustees. REEMPLOYMENT SERVICES, TRAINING, EDUCATION, AND INCOME SUPPORT UI recipients who are not on temporary layoff may benefit from reemployment services designed to help them find a job. These services could include referrals to job openings, training in job search techniques, help with resumes, provision of information about jobs in demand, occupational aptitude and interest testing, and other similar assistance. Reemployment services may help recipients find jobs more quickly and may lead to better job matches, and recipients with poor labor market prospects might have strong incentives to use these services. Occupational training or further education may help some recipients with weak or outmoded job skills improve their skills, allowing them to find better jobs than they otherwise would. Despite the poor labor market outcomes of some recipients and recent innovations to improve reemployment service delivery, we find: vi
Recipients in 1998 were less likely than in 1988 to use reemployment services. Forty-one percent in 1998, as compared to 54 percent in 1988, contacted the Job Service or a one-stop career center shortly after beginning their UI claim. Some recipients who contacted the Job Service or one-stop reported not receiving any services, and the prevalence of nonreceipt has increased over time. This was true for 37 percent of recipients who contacted the Job Service or a one-stop in 1998, compared to 28 percent in 1988. Recipients who did not receive specific services probably registered with the Job Service and attended an orientation session on available services but did not use them. Recipients in 1998 who received additional services received 2.1 on average, with a job referral being the most common. A substantial portion of services in 1998 were provided through self-access resource centers. Based on the survey data, about 35 percent of recipients said that they received a notice requiring them to report to the Job Service or a one-stop. Most of these callin notices were probably generated by the Worker Profiling and Reemployment Service (WPRS) systems that states implemented in the mid-1990s to direct services toward likely exhaustees. Information on the characteristics of these recipients suggests that, to some degree, states are successfully targeting services to likely exhaustees. About three-quarters of the recipients who received these call-in notices said they went for services. This group received more services than other recipients who went to the Job Service or a one-stop. Most recipients who participated in training or education entered programs designed to improve their occupational skills. The rate of participation in training or education programs was somewhat higher in 1998 than in 1988 (14 versus 11 percent). By the time of the interview, most people in these programs had completed their training or education or were still participating. Most of them considered this training or education helpful in obtaining a job and useful on the jobs they held. UI recipients families experienced large declines in weekly income during their unemployment spells relative to their pre-unemployment situations. UI benefits provided an important source of income during this time, as did the earnings of spouses or partners. However, this latter source of income was only available to about 40 percent of the UI recipients families. The remaining families relied almost solely on UI benefits for income support. Relatively few UI recipients or their families received income from retirement benefits, welfare, or other transfer programs. INTERPRETATION AND IMPLICATIONS FOR POLICY The labor market of the late 1990s was one of the strongest of the postwar era, yet the labor market outcomes reported in this study for UI recipients, and especially for exhaustees, are surprisingly poor. UI recipients in 1998 were both less likely to have a job two years after their initial job separations and took longer to become reemployed when they did so than were UI vii
recipients in 1988. As in earlier periods, exhaustees experiences in 1998 were worse than those of other UI recipients more than one-fourth of exhaustees never had a job in the post-ui period, and of those who did find employment, 30 percent had earnings at least 25 percent lower than they had before becoming unemployed. Clearly, many recipients were left behind in the highpressure labor market of the late 1990s. Despite the fact that UI recipients in 1998 were having difficulty finding jobs, they were less likely than recipients in 1988 to seek reemployment services from the Job Service or a one-stop career center. This reduction in use of reemployment services occurred both at the start of UI collection and following benefit exhaustion. Why were UI recipients labor market outcomes so poor? There are two plausible interpretations of the general finding that UI recipients fared poorly in the late 1990s: (1) the strength of the overall labor market permitted most workers to avoid collecting UI, which caused the pool of 1998 UI recipients to include a disproportionate number of workers with significant labor market problems; or (2) factors in the overall labor market changed such that UI recipients faced new difficulties that were not as prevalent in the past. Although we cannot unambiguously distinguish between these two possibilities, evidence from this study that shows that nonexhaustees as well as exhaustees experienced relatively poor labor market outcomes. Other research shows that unemployment durations are increasing and that dislocation is affecting a broader spectrum of workers. These findings lead us to conclude that important changes in labor markets have occurred that, in part, explain our results. UI policymakers will need to monitor such changes and study their programmatic implications. Why did the use of reemployment services decline? We investigated a number of potential explanations for this finding and conclude that a combination of factors probably contributed to this outcome. The implementation of WPRS systems, which were intended to direct services to likely exhaustees, may have concentrated services on a smaller group of recipients than in the past. A reduction in the capacity of the Job Service to provide services and recipients reactions to a strong labor market may also have played a role, but in each case the evidence is mixed. The introduction in some states of remote UI initial claims processing might have also had an effect, since recipients no longer need to go to local offices to file claims, but the decline in service use occurred more widely than the introduction of remote claims. It is unlikely that changes in the composition of recipients affected service use, since changes that would decrease service use were offset by changes that would increase service use. How might labor market outcomes be improved? In the past, the UI system has used a combination of job search requirements and reemployment services to promote rapid reemployment of recipients. Strengthening job search requirements and/or increasing the resources available for reemployment services would promote this objective. Improving the targeting of the current level of resources might also be appropriate. Simulations of alternative targeting mechanisms suggest that: viii
- Services are currently targeting expected exhaustees. Simulations showed that services were indeed targeting UI recipients who were likely exhaustees and that these recipients were experiencing serious labor market difficulties. - Services are currently targeting dislocated workers. Simulations suggested that dislocated workers were much more likely to obtain various reemployment services than were other UI recipients. Labor market outcomes for dislocated workers were also significantly worse than for other workers. - Low-skill recipients may have unmet needs for services. Simulations showed that low-skill UI recipients were no more likely than other UI recipients to obtain reemployment services. However, their labor market outcomes were significantly worse than those of other recipients, suggesting that this group is underserved. ix